Today we performed some investigations in order to figure out the excess noise in MICH and its possible causes:
-
for the first part of the shift we focused on the B4 signals, the one coming from SSFS and the standard SPRB one:
- we tried to use both the B4_56MHz_I signals for the SSFS, the SPRB one for the low frequency branch (IMC) and the usual SSFS one for the high frequency branch; however the switch for the low frequency region did not work. This could be due to the different structures which appear in the two signals, who were after all one of the signs of something going on with the B4 sensor; the TF between the two spans over different orders of magnitude in the whole frequency range, so a straightforward way to to the hand-off is not obvious; a follow-up test could be to try the hand-off before the SSFS boost is engaged, as the difference between the signals is less significant in that case, and then engage the SSFS boost;
- as a secondary trial we switched the MICH sensor, from the SSFS B4_56MHz_Q signal to its SPRB counterpart; the TF between the two was flat as already measured before, and the hand-off was consequently smooth, but no changes were seen on MICH itself or on the other B4 sensors/quadratures;
-
at this point we investigated other error signals for the MICH loop, in order to try and find if the same structures appeared with other sensors; in the end we switched MICH to B2_56MHz_Q and we compared several configurations for MICH, which are reported in the Figure:
- MAGENTA is the usual MICH sensor in dark fringe;
- BLACK is the simple Michelson in DC readout, which is well below the usual spectrum and with most of the structures not present;
- GREEN is MICH on B2_56MHz_Q: what is striking here is the much higher bump in the low frequency region; it is unclear if this means that B2_56MHz is just a worse sensor than B4_56MHz_Q, or if they are actually sensing something different and the B4 signal is in some way affected by the SSFS loop;
- RED and BLUE are the error signal and the correction for the B2_56MHz_Q lock;
- During the shift the B4 demodulation phases moved again a little w.r.t. this morning, and the SSFS phase offset is still not linearly added to the SPRB phase for the SSFS signals;
Summing up, there is no clear conclusion yet on the source of this noise: next steps could be working more on the two branches of the SSFS using the different sensors, looking at different MICH configurations at dark fringe, or study the MICH sensors in different/simpler configurations (Michelson, CITF).
While I was trying to restore the ITF for the night, I found that it got stuck in the FmodErr sequence, often looping between the different FmodErr states (THERMAL_CORRECTION, LNFS_TUNING and MC_TUNING); after selecting manually IMC_RESTORED I left the ITF in LOCKED_RECOMBINED hoping not to trigger the loop again during the night.